Cyclopamine Modulates γ-Secretase-mediated Cleavage of Amyloid Precursor Protein by Altering Its Subcellular Trafficking and Lysosomal Degradation

Background: Sterols can alter APP metabolism.

Results: Cyclopamine, a phytosterol, alters APP-CTF degradation rate, decreases APP-CTF bioavailability for γ-secretase cleavage, and reduces Aβ and AICD generation.

Conclusion: Cyclopamine decreases Aβ and AICD production by altering APP-CTF retrograde trafficking.

Significance: Cyclopamine is a novel modulator of APP metabolism and trafficking, which can illuminate new avenues for Alzheimer disease treatment.

Abstract

Alzheimer disease (AD) is a progressive neurodegenerative disease leading to memory loss. Numerous lines of evidence suggest that amyloid-β (Aβ), a neurotoxic peptide, initiates a cascade that results in synaptic dysfunction, neuronal death, and eventually cognitive deficits. Aβ is generated by the proteolytic processing of the amyloid precursor protein (APP), and alterations to this processing can result in Alzheimer disease. Using in vitro and in vivo models, we identified cyclopamine as a novel regulator of γ-secretase-mediated cleavage of APP. We demonstrate that cyclopamine decreases Aβ generation by altering APP retrograde trafficking. Specifically, cyclopamine treatment reduced APP-C-terminal fragment (CTF) delivery to the trans-Golgi network where γ-secretase cleavage occurs. Instead, cyclopamine redirects APP-CTFs to the lysosome. These data demonstrate that cyclopamine treatment decreases γ-secretase-mediated cleavage of APP. In addition, cyclopamine treatment decreases the rate of APP-CTF degradation. Together, our data demonstrate that cyclopamine alters APP processing and Aβ generation by inducing changes in APP subcellular trafficking and APP-CTF degradation.

Introduction

Alzheimer disease (AD)³ is the most common form of dementia. It is an irreversible neurodegenerative disease characterized by gradual cognitive decline (1, 2). AD is neuropathologically characterized by senile plaques, composed of amyloid-β (Aβ), and neurofibrillary tangles, composed of hyperphosphorylated Tau. The amyloid cascade hypothesis suggests neurotoxic Aβ initiates a series of events that result in synaptic dysfunction leading to neuronal loss (3–9). The Aβ peptide is produced by proteolysis of amyloid precursor protein (APP) (10–15).

APP is a type I transmembrane protein containing an intracellular C-terminal domain and a larger external N-terminal domain (16, 17). APP is proteolytically processed via nonamyloidogenic or amyloidogenic pathways. After translation, N- and O-glycosylated APP (mature APP) is trafficked to the plasma membrane where α-secretase (nonamyloidogenic) can cleave full-length APP (FL-APP) liberating a soluble N-terminal fragment (sAPPα) and a membrane tethered C-terminal fragment (APP-CTFα). Alternatively, FL-APP can be endocytosed and either recycled back to the plasma membrane or localized to the early endosome where β-secretase cleavage of FL-APP initiates amyloidogenic cleavage. This β-secretase cleavage liberates a soluble N-terminal fragment (sAPPβ) and a membrane-tethered C-terminal fragment (APP-CTFβ) (18). Following this initial cleavage by either α- or β-secretase, retrograde trafficking mechanisms deliver APP-CTFs to the trans-Golgi network (TGN). Recent literature demonstrates that the majority of γ-secretase cleavage of APP occurs at the TGN (19). Here, APP-CTFs are cleaved by γ-secretase to generate the APP intracellular domain (AICD) and p3 or AICD and Aβ depending on whether the APP-CTF was derived from α- or β-secretase, respectively.

APP proteolysis is intimately associated with its subcellular localization; therefore, APP trafficking plays a critical role in amyloidogenesis. Cholesterol can alter cellular membrane fluidity and therefore can alter trafficking of transmembrane proteins like APP (20, 21). Cholesterol has been shown to play an important role in AD risk and pathogenesis (22–24). High cholesterol levels have been implicated as an AD risk factor (25–27). Clinically, patients treated with statins, inhibitors of cholesterol synthesis, have reduced risk for AD (23, 24, 28). Increasing cholesterol levels in in vitro and in vivo AD models exacerbates Aβ production (25, 29, 30). Conversely, inhibiting cholesterol synthesis in vitro or in vivo reduces Aβ generation (31).

Recently, phytosterols were also demonstrated to modulate Aβ generation. Stigmasterol treatment decreased Aβ generation by modulating γ-secretase activity and β-secretase trafficking (32). Cyclopamine is a naturally occurring plant phytosterol from the corn lily (Veratrum californicum) plant (33). Cyclopamine and its analogs have been used to treat cancer, specifically, in combination with lovastatin to treat medulloblastoma (34, 35).

Here, we demonstrate that cyclopamine modulates APP metabolism. Specifically, cyclopamine prevents Aβ generation by decreasing γ-secretase-mediated cleavage of APP-CTFs. This occurs due to alteration in subcellular trafficking of APP-CTFs. Upon cyclopamine treatment, APP-CTFs accumulate in lysosomes rather than being trafficked to the TGN. Once in the lysosomes, APP-CTF degradation is decreased leading to its accumulation. Together, our data demonstrate a novel use of cyclopamine and may open new avenues to treat Alzheimer disease.

EXPERIMENTAL PROCEDURES

Antibodies, Plasmids, and Reagents

Antibodies were obtained from the following: mouse C-terminal APP clone c1/6.1 (a kind gift from P. Mathews, Nathan Kline Institute, New York); mouse APP N-terminal 4A 22C11 (Millipore); rabbit APP C-terminal A8717, mouse anti-Myc clone 9E10, rabbit LC3IIB, and mouse β-actin (Sigma); mouse cleaved Notch1 clone D3B8 and rabbit PSEN1 CTF clone D39D1 (Cell Signaling); mouse LAMP1, TGN38, and mouse EEA1 (BD Biosciences); rat LAMP1 clone 1D4B and mouse MP6R clone 22d4 (DSHB); rabbit TGN46 (AbD Serotec), and rabbit cathepsin D (kind gift from Dr. Stefan Höning). Fluorescent secondary antibodies (Alexa Fluor 488 and 594) were from Jackson ImmunoResearch and IR-conjugated secondary antibodies (IRDye680 and IRDye800) were from Li-Cor Biosciences. Peroxidase-conjugated secondary antibodies were from Cell Signaling. Cyclopamine (0.5–10 μm) was purchased from LC Laboratories, and L-685,458 (2 μm), cycloheximide (50 μg/ml), and DMSO were from Sigma. pCS2-Myc-ΔENotch was used for overexpression studies. pMst-APP-Gal4 was originally developed by Cao and Sudhof (36). Human Chmp2a-GFP construct was a gift from Dr. Elias Spiliotis.

Cell Culture and Transfection

HeLa cells were maintained at 37 °C, 5% CO₂ in complete DMEM (Corning Glass) supplemented with 10% FBS (Atlanta Biologicals), 100 units/ml penicillin, and 100 μg/ml streptomycin (Corning Glass), 2 mm l-glutamine (Corning Glass). Cells were grown to 80% confluence and serum-starved (0.5% FBS/DMEM) for 24 h prior to pharmacological or genetic manipulation. For pharmacological manipulation, drugs were diluted in 0.5% FBS/DMEM. For genetic overexpression experiments, cells were grown to 80% confluence and then transfected using with TurboFect transfection reagent (Thermo Scientific) according to the manufacturer's protocol. Culture media were removed 24 h post-transfection, and cells were collected or further treated with pharmacological agents (0.5% FBS/DMEM) for an additional 24 h.

Primary Neuron Culture

Primary cortical neuron cultures were isolated from postnatal day 1 (P1) Sprague-Dawley rat pups. Briefly, cortices were dissected out, minced, and treated with papain (100 units; Worthington) for 15 min at 37 °C. Next, tissue was treated with type II-O trypsin inhibitor from chicken egg white (Sigma) for 15 min at 37 °C. Tissue was washed with fresh Neurobasal medium (Invitrogen) supplemented with B-27 (Invitrogen), 2 mm l-glutamine, 100 units/ml penicillin, and 100 μg/ml streptomycin. Tissue was triturated, centrifuged at 1000 rpm for 10 min, and then resuspended in the fresh and complete Neurobasal medium. 2 × 10⁶ cells per 35-mm well were plated onto poly-dl-lysine (50 μg/ml; Sigma)-coated tissue culture plates. Cortical neurons were treated with pharmacological agents on 6 days in vitro for 24 h, and lysates were collected for further biochemical analysis. All animals were used in accordance with animal protocols approved by the Institutional Animal Care and Use Committee (IACUC Protocol number 19787). Animals were delivered to and maintained at the Calhoun Animal Facility (Drexel University, PA). Animal procedures were performed strictly in accordance with the National Institutes of Health Guide for the care and use of Laboratory Animals approved by the Drexel University Animal Care and Use Committee.

Drosophila Stocks and Genetics

Drosophila husbandry was performed as described previously (37). For experiments utilizing the γ-secretase reporter GMR-APP-Gal4; UAS-Grim/Cyo model (38), flies were crossed on standard cornmeal agar food supplemented with cyclopamine (100 nm) or DMSO vehicle control (0.1%), and flies were collected 24 h post-eclosion, and their compound eye was inspected. Assessment of penetrance and severity of the rough-eye phenotype was accomplished by photographing the compound eye using a Canon PowerShot S70 digital camera mounted to a Leica Mz 125 stereomicroscope. Severity of rough-eye phenotype was scored + (mild) to +++ (severe). One “+” refers to where less than ½ of the eye was apoptotic and therefore appears “rough”. A score of “++” (moderate) defined increased penetrance, where apoptosis affected approximately ½ of the eye. Severe “+++” rough-eye phenotype described when more than ½ of the eye appeared rough, and eye size was significantly reduced. For objective quantification, five blinded laboratory personnel analyzed all experiments.

Immunoblotting

Lysates were collected in complete RIPA buffer (50 mm Tris-HCl, pH 8.0, 150 mm NaCl, 0.5% sodium deoxycholate, 0.1% SDS, 1% Nonidet P-40) supplemented with Halt protease and phosphatase inhibitor and EDTA (ThermoFisher). Lysates were briefly cleared at 20,000 × g at 4 °C and stored at −20 °C. Protein concentrations were determined using the BCA assay kit according to the manufacturer's protocol (Pierce). 40 μg of lysate was supplemented with NuPAGE LDS sample buffer (Invitrogen) and heated to 75 °C for 10 min. Protein was separated on 4–12% NuPAGE BisTris gels (Invitrogen) using MES running buffer (Invitrogen) and then transferred onto Immobilon PVDF membrane (Millipore). Odyssey blocking buffer (Li-Cor Biosciences) was used for blocking and resuspending primary and secondary antibodies. Membranes were scanned using Li-Cor Odyssey infrared scanning instrument.

Aβ ELISA

HeLa cells and primary rat cortical neurons were treated with pharmacological agents for 24 h, and conditioned supernatants were collected and cleared at 20,000 × g for 20 min at 4 °C. Fresh cleared supernatants were used for Aβ₄₀ ELISA kit (Wako, Japan) according to the manufacturer's protocol. Briefly, samples were diluted 1:1 using kit diluent and incubated overnight at 4 °C. Samples were compared with the ELISA kit positive control and negative control (diluent alone). Samples were incubated and analyzed using a luminescence plate reader.

In Vitro γ-Secretase Assay

We utilized a well established cell-free γ-secretase activity assay that utilizes a fluorogenic peptide substrate corresponding to the APP γ-secretase cleavage site (39, 40). HeLa cells grown to 100% confluence in 150-mm culture dishes were collected in ice-cold PBS and pelleted at 5000 rpm for 5 min. The pellet was homogenized in 500 μl of Buffer B (20 mm HEPES, pH 7.5, 150 mm KCl, 2 mm EGTA, protease and phosphatase inhibitors) using a 27-gauge needle. The resulting homogenate was cleared at 45,000 rpm at 4 °C for 1 h. Supernatant was stored at −80 °C, and pellet was washed with 500 μl Buffer B and passed through 27 gauge needle on ice. The suspension was cleared again at 45,000 rpm for 1 h at 4 °C. Supernatant was discarded and pellet resuspended in 75 μl Buffer B + 1% CHAPSO and passed through a 27-gauge needle on ice. The resulting membrane samples were solubilized on a rotator at 4 °C for 2 h. Solubilized samples were cleared at 45,000 rpm for 1 h at 4 °Cl; supernatant (total cell membrane) was collected and pellet discarded. Total protein was determined using BCA assay (Pierce) and 200 μg of protein were used for in vitro γ-secretase activity assay. Briefly, membranes were resuspended in γ-secretase assay buffer (100 mm Tris-HCl, pH 6.8, 4 mm EDTA, 0.5% CHAPSO), and pretreated with vehicle control, L-685,458, or cyclopamine. Because the membrane preparation enriches total γ-secretase in the sample, the amount of pharmacological agent was increased accordingly. Therefore, 20 μm drug in a total vehicle volume of 1 μl per was used. 150 μl of total volume per well of a 96-well plate was used. Membranes were pretreated for 3 h at 37 °C and 5% CO₂, and then fluorogenic γ-secretase substrate (Calbiochem, EMD Millipore) was added to membranes and further incubated at 37 °C and 5% CO₂ for the indicated time points at which time membranes were removed and fluorescence was measured using a plate reader (Promega). BSA was used as negative control in place of membranes.

Subcellular Fractionation

HeLa cells grown to 80% confluence in 100-mm culture dishes were treated with 5 μm cyclopamine or DMSO for 24 h, rinsed and collected in PBS, and then cleared at 1000 rpm for 7 min. The cell pellet was resuspended in homogenization buffer (250 mm sucrose, 150 mm NaCl, 25 mm Tris, 1 mm EDTA, protease and phosphatase inhibitor mixture) and homogenized using ball-bearing 12-μm clearance cell buster. Homogenates were cleared at 1000 × g for 15 min at 4 °C, and post-nuclear supernatant was loaded into discontinuous density gradient (50, 30, and 10%) medium (Optiprep, Sigma) in Opti-Seal centrifuge tubes (Beckman). Homogenates were spun at 30,000 rpm for 19 h at 4 °C and 300 μl fractions collected.

Immunofluorescence

Cells were fixed using 4% paraformaldehyde, 0.1% Triton X-100, blocked in 2% BSA for 30 min, and incubated with primary antibodies overnight at 4 °C. Cells were rinsed with PBS and stained with secondary antibodies at room temperature for 1 h, washed with PBS, and mounted. Cells were imaged using Olympus Fluoview 1000 inverted confocal microscope. Quantification of three-dimensional confocal image stacks was accomplished using SlideBook 5.0 or Volocity Image analysis software (PerkinElmer Life Sciences).

Surface Biotinylation

HeLa cells were treated with 5 μm cyclopamine or DMSO for 24 h. Cells were placed on ice to halt membrane dynamics, rinsed with ice-cold PBS, and incubated with sulfo-NHS-SS-biotin (1 mg/ml in PBS; Thermo Scientific) for 40 min on ice with gentle rocking. Biotin was quenched with 100 mm glycine in PBS for 15 min. Cells were collected in PBS and pelleted at 500 × g for 5 min at 4 °C. The pellet was lysed in 200 μl of standard RIPA lysis buffer containing protease and phosphatase inhibitors. Lysate was sheared using a 27-gauge needle on ice and solubilized for 2 h at 4 °C on a rotator. Lysate was cleared by centrifugation at 10,000 × g for 5 min, and 50 μl from each sample was set aside for “total” protein analysis. The rest of the supernatant was loaded into a capped spin column (Pierce; 69725) with NeutrAvidin-coated agarose resin (Thermo Scientific) at a 1:1 ratio and incubated overnight at 4 °C on a rotator. Columns were centrifuged at 10,000 rpm for 1 min, and flow-through (“unbound” control) was collected and saved. Resin was washed several times with complete RIPA. Then 50 μl of NuPAGE LDS Sample Buffer (Invitrogen) with 5% β-mercaptoethanol was loaded into each column and incubated for 30 min at room temperature on a shaker. To collect the surface-biotinylated protein, columns were centrifuged at maximum speed for 2 min. Biotinylated protein was separated on 4–12% NuPAGE BisTris gels (Invitrogen) and then transferred, and the membrane was probed for surface APP. Nonbiotinylated lysates were collected as control samples. Biotinylated “surface” samples were compared with total lysate samples.

Statistical Analysis

All graphs and diagrams represent mean values ± S.E. of all triplicates from at least three independent experiments. Either two-tailed or one-tailed Student's t test was used to compare two treatment groups and calculate significance from at least three independent experiments (*, p < 0.05; **, p < 0.01; ***, p < 0.005). For in vivo Drosophila melanogaster experiments, G test (goodness of fit) was used to determine significance of phenotypic penetrance in experimental populations. Degree of significance and corresponding p value criteria for G test were identical to previously mentioned Student's t test.

RESULTS

Cyclopamine Treatment Results in APP C-terminal Fragment Accumulation

To test whether cyclopamine modulates APP metabolism, we treated primary rat cortical neurons with cyclopamine (41, 42). We did not observe an appreciable change in the full-length APP (FL-APP) holoprotein after 24 h of 5 μm cyclopamine treatment (Fig. 1, A and B). However, the 8–12-kDa APP products of α- and β-secretase (α- and β- CTFs; collectively known as APP-CTFs) significantly increased when compared with vehicle control-treated neurons (p = 0.0190) (Fig. 1, A and C). To determine whether these effects can be observed in other models, we utilized HeLa cells because they are easily manipulated and have been previously utilized to study APP processing and trafficking (19, 43). Using naive HeLa cells, we performed cyclopamine time and dose dependence experiments. Cells were treated for 24 h with increasing concentrations of cyclopamine from 0.5 to 10 μm (Fig. 1D). Compared with vehicle control (Fig. 1E, dashed line), we observed a significant increase in APP-CTF levels with as little as 0.5 μm cyclopamine (p = 0.000615) (Fig. 1E). No change in FL-APP was observed in cells exposed to 0.5, 1, and 5 μm of drug (Fig. 1E). A small, yet significant increase in FL-APP was observed upon 10 μm cyclopamine treatment. To address time dependence of cyclopamine's effects on APP-CTF accumulation, we performed a time course experiment. Because we observed robust increases in APP-CTFs after 24 h with as little as 0.5 μm drug, we hypothesized that using 5 μm cyclopamine would significantly increase APP-CTF levels within a shorter exposure time. Accumulation of APP-CTFs was evident by 3 h of exposure (p = 0.000488), and further accumulation continued for the remainder of the time course (by 24 h p = 9.50 × 10⁻⁶) (Fig. 1, F and G). The lack of significant changes in FL-APP levels upon 5 μm cyclopamine exposure suggests APP gene transcription is not altered. In fact, using quantitative PCR, we analyzed APP mRNA in naive HeLa cells upon cyclopamine treatment and did not observe changes in APP transcript levels as compared with vehicle control (data not shown).

FIGURE 1.

Cyclopamine treatment alters APP metabolism. A, primary rat cortical neurons treated with 5 μm cyclopamine (Cyc) for 24 h. Endogenous FL-APP and APP-CTFs were detected using Western immunoblotting and a C-terminal APP antibody (c1/6.1). B and C, FL-APP protein levels and APP-CTFs were normalized to β-actin and FL-APP, respectively. D, HeLa cells treated with the indicated concentrations of cyclopamine for 24 h followed by FL-APP and APP-CTF detection in cell lysates via Western immunoblotting. E, quantification of dose-dependent accumulation of APP-CTFs normalized to FL-APP. Relative protein changes were compared with vehicle (DMSO) control (E, dashed line). F, Western blot time course analysis of endogenous FL-APP and APP-CTFs in HeLa cells treated with 5 μm cyclopamine. G, normalized APP-CTFs increase in a time-dependent manner as compared with vehicle (DMSO) control (G, dashed line). H, Using Western blot analysis and N-terminal APP antibody (22C11), endogenous sAPP levels were monitored in supernatants of HeLa cells treated with 5 μm cyclopamine for 24 h. I, sAPP levels were normalized to protein concentrations in lysates, determined using BCA assay. Lack of change in sAPP protein levels are illustrated in comparison with vehicle control (DMSO). J, endogenous FL-APP and APP-CTFs from naive HeLa cells treated with 5 μm cyclopamine were compared with cells treated with 2 μm L-685,458 using Western immunoblot analysis and C-terminal APP antibody (c1/6.1). Lower panel represents increased exposure of APP-CTFs. K and L, cyclopamine increased APP-CTF levels (not FL-APP) compared with vehicle (DMSO) control. Values denote mean ± S.E. Student's t test was used for statistical analysis: ***, p < 0.005; **, p < 0.01; *, p < 0.05.

APP proteolysis is initiated by α- or β-secretase. This cleavage liberates soluble N-terminal APP ectodomains (sAPP). Treatment of naive HeLa cells with cyclopamine did not alter sAPP levels (Fig. 1, H and I). This suggests the increase in APP-CTFs is not due to modulation of α- or β-secretase cleavage of APP by cyclopamine. The observed increase in APP-CTFs and the lack of change in FL-APP levels resemble the effects of γ-secretase inhibitors but to a diminished degree (Fig. 1, J–L) (44, 45).

Cyclopamine Decreases γ-Secretase-mediated Cleavage of APP in Vitro and in Vivo

Because cyclopamine treatment increased levels of APP-CTFs, analogously to γ-secretase inhibitor treatment, we hypothesized that cyclopamine would decrease levels of γ-secretase cleavage products, namely Aβ and the AICD. Both Aβ and AICD are produced upon γ-secretase cleavage of APP-CTFs. To test this hypothesis, we exposed naive HeLa and primary rat cortical neuron cells to cyclopamine for 24 h. Cyclopamine-treated cells secreted significantly less Aβ compared with vehicle control in primary cortical neurons and HeLa cells (p = 0.00567) (Fig. 2A). The other product of γ-secretase cleavage, AICD, is difficult to detect. Therefore, we used a previously described APP-Gal4 construct to aid in detection (36, 46). We exposed HeLa cells transiently overexpressing APP-Gal4 to cyclopamine for 24 h and detected AICD-Gal4 levels using Western blot analysis. The γ-secretase inhibitor L-685,458 served as a positive control because it prevents AICD generation. As predicted, in comparison with vehicle control, cyclopamine significantly decreased AICD-Gal4 levels (p = 0.000228) (Fig. 2, B and C). However, these effects were much more modest than those observed upon γ-secretase inhibition with L-685,458.

FIGURE 2.

Cyclopamine decreases γ-secretase cleavage of APP in vitro and in vivo. A, secreted Aβ₄₀ levels from HeLa cells and primary cortical neurons (6 days in vitro) treated with 5 μm cyclopamine (Cyc) for 24 h were measured using an ELISA and compared with vehicle (DMSO) control. Secreted Aβ₄₀ levels were normalized to total protein in respective cell lysates. B, HeLa cells transiently overexpressing FL-APP-Gal4 treated with vehicle (DMSO) control, 2 μm L-685,458, or 5 μm cyclopamine for 24 h. APP-CTF-Gal4 and AICD-Gal4 levels analyzed using Western immunoblotting and a C-terminal APP antibody (c1/6.1). C, APP-AICD-Gal4 levels normalized to APP-CTF-Gal4 levels. D, HeLa cells transiently overexpressing ΔENotch-Myc treated with vehicle (DMSO) control, 2 μm L-685,458, or 5 μm cyclopamine for 24 h. NICD and ΔENotch levels were analyzed using Western immunoblotting and anti-Myc and anti-NICD (Val-1744) antibody. E, NICD levels normalized to ΔENotch levels. F, APP-CTF-Gal4 and ΔENotch levels normalized to β-actin. G–J, representative images of rough-eye phenotype from GMR-APP-Gal4; UAS-Grim (γ-secretase reporter) flies raised on normal food, 100 nm cyclopamine, or vehicle control (DMSO). Flies were scored 1 day after eclosion as follows: mild (+), moderate (++), and severe (+++). K, relative changes in penetrance of rough-eye phenotype in flies raised on normal food (n = 168), vehicle-containing food (n = 72), or cyclopamine-containing food (n = 112). Population of flies with mild, moderate, or severe rough-eye phenotype is illustrated as percent of total population per experimental group. Statistical analysis in vivo experiments are as follows: G test, ***, p < 0.005; **, p < 0.01, was performed. Student's t test was used for statistical analysis of cell based in vitro studies. Values denote means ± S.E. n.s., not significant.

To determine whether the observed effects were specific to APP, we monitored γ-secretase cleavage of Notch in response to cyclopamine treatment. Similar to AICD, endogenous NICD is also difficult to detect. To overcome this difficulty, we transiently overexpressed Myc-ΔENotch in HeLa cells and treated with cyclopamine for 24 h (47, 48). Cyclopamine significantly decreased NICD levels (p = 8.87 × 10⁻⁵) to a similar extent as observed in Aβ and AICD levels (Fig. 2, D and E). These effects on AICD and NICD were much more modest than those observed upon γ-secretase inhibition. Similar to endogenous APP-CTFs, cyclopamine increased APP-CTF-Gal4 levels (Fig. 2F). Interestingly, no change in ΔENotch levels was observed suggesting that the effects could be specific (Fig. 2F).

Given these results and the availability of an in vivo γ-secretase reporter, we tested the ability of cyclopamine to modulate γ-secretase cleavage of APP in vivo. Briefly, in 2003, Guo et al. (38) developed and characterized a D. melanogaster γ-secretase reporter. These transgenic flies express the APP γ-secretase substrate, APP-C99-Gal4, specifically in the fly eye ommatidia. These flies also carry a UAS element upstream of GRIM, a cell death activator. Upon γ-secretase cleavage of APP-C99-Gal4, the resulting AICD-Gal4 can bind to the UAS element and induce GRIM expression. GRIM expression leads to death of ommatidia and results in a rough-eye phenotype (38, 49, 50). To test whether cyclopamine decreases γ-secretase-mediated cleavage of APP-C99-Gal4, we raised APP-C99-Gal4;UAS-GRIM flies on normal, vehicle, or cyclopamine supplemented food. Flies were collected 1 day post-eclosion, and their eyes were scored for rough-eye phenotype. Flies raised on cyclopamine displayed decreased severity of the rough-eye phenotype (p = 2.40 × 10⁻³⁴) (Fig. 2, G–K). More specifically, 10% of the flies raised on cyclopamine displayed severe rough-eye phenotype compared with the 47% raised on vehicle food. Although only 10% of the vehicle-treated flies displayed “mild” rough-eye phenotypes, 53% of cyclopamine-treated flies displayed this phenotype (Fig. 2K). Together, these data demonstrate that cyclopamine treatment decreases γ-secretase-mediated cleavage of APP-CTFs in vitro and in vivo.

Cyclopamine Does Not Alter γ-Secretase Activity

Because in vivo and in vitro cyclopamine treatment leads to decreased γ-secretase cleavage of APP-CTFs, we investigated whether cyclopamine inhibits γ-secretase activity. One major step in γ-secretase complex maturation is the autoproteolysis of presenilin1 (PSEN1) to form the active N- and C-terminal fragments. Therefore, detection of the PSEN1-CTF is an indicator of an active γ-secretase complex (11). To this end, we exposed naive HeLa cells to cyclopamine for 24 h and observed an increase in APP-CTFs levels; however, PSEN1-CTF levels did not change in response to cyclopamine treatment (Fig. 3, A and B).

FIGURE 3.

γ-Secretase activity is not altered by cyclopamine treatment. A, HeLa cells treated with 5 μm cyclopamine (Cyc) or vehicle control (DMSO) for 24 h. PSEN1-CTF levels were from respective lysates analyzed via Western immunoblotting. B, PSEN1-CTFs levels normalized to β-actin. C, fluorometric γ-secretase activity assay. Fluorescence intensity over time using total membranes isolated from naive HeLa cells treated with cyclopamine (20 μm), L-685,458 (20 μm), or vehicle control (DMSO). Graph represents relative changes in fluorescence as percent activity of control (membranes treated with DMSO) over time.

To assess overall γ-secretase activity, we utilized an in vitro, fluorescence-based activity assay (40). We isolated total cellular membranes from naive HeLa cells and treated these membranes with vehicle, L-685,458, or cyclopamine (51–54). As expected, treatment with L-685,458 decreased cleavage of the fluorogenic γ-secretase peptide substrate resulting in decreased fluorescence intensity over time. Surprisingly, treatment with cyclopamine did not alter γ-secretase activity (Fig. 3C). These results suggest that cyclopamine decreases γ-secretase-mediated cleavage of APP without directly affecting γ-secretase activity. One mechanism that could explain these results is that cyclopamine mediates a change in the subcellular localization of APP and/or γ-secretase.

Cyclopamine Alters APP-CTF Subcellular Localization

Proteolytic processing of APP is dependent on its subcellular localization. To investigate whether cyclopamine alters APP subcellular localization, naive HeLa cells were exposed to cyclopamine for 0, 6, or 24 h, and APP subcellular distribution was visualized using immunofluorescence. Analogous to the time course experiment in which we observed increased APP-CTFs by Western blot (Fig. 1G), here cyclopamine treatment induced significant accumulation of APP-positive puncta detected with an antibody raised to the APP C terminus (p = 0.00120) (Fig. 4, A–C). Visualization of APP distribution using an antibody specific to the N-terminal portion of APP did not reveal similar cyclopamine-induced APP puncta. In fact, a lack of colocalization was observed between the N- and C-terminal APP antibodies in the cyclopamine-induced APP puncta (Fig. 4, D and E). This suggests that the cyclopamine-induced puncta are APP-CTFs and not FL-APP nor sAPP. FL-APP is not a suitable substrate for γ-secretase cleavage. The increase in APP-CTF subcellular puncta and the lack of change in FL-APP and sAPP protein levels suggests that cyclopamine does not alter APP biosynthetic pathway. Furthermore, it also suggests that cyclopamine may induce alterations in APP-CTF endocytosis.

FIGURE 4.

Cyclopamine induces subcellular accumulation of APP-CTF. Confocal three-dimensional analysis of HeLa cells treated with vehicle control (DMSO) (A) or 5 μm cyclopamine (B) for 0, 6, or 24 h. APP was detected using an antibody against the C terminus of APP (A8717). Scale bar, 10 μm. C, number of APP puncta normalized to regions of interest (ROI). Regions of interest = 10 × 10 μm, and 20–30 regions of interest per treatment were analyzed. Puncta were defined as 2× the intensity of background in the cytosolic nonpunctate region, and objects were restricted to 0.2–2.0 μm skeletal diameter. Cyc, cyclopamine. Confocal three-dimensional analysis of naive HeLa cells treated with vehicle control (DMSO) (D) or 5 μm cyclopamine (E) for 24 h. Cells were stained using C-terminal APP (A8717) (left panel) and N-terminal APP (22C11) (middle panel) antibodies. Right panel denotes merged channels. Scale bar, 10 μm. F, cell surface FL-APP levels in HeLa cells treated with 5 μm cyclopamine or vehicle (DMSO) for 24 h and then biotinylated. Biotinylated surface FL-APP levels were measured by Western blot with a C-terminal APP antibody (c1/6.1). Top panel is total FL-APP from whole cell lysate, middle panel is biotinylated surface FL-APP purified with NeutrAvidin-coated resin. G, surface FL-APP levels normalized to total FL-APP as percent of vehicle (DMSO) control. Values denote means ± S.E. Student's t test was used for statistical analysis; **, p < 0.01.

To further investigate this latter possibility, we measured surface APP levels using cell surface biotinylation and Western blot analysis. We treated naive HeLa cells with cyclopamine for 24 h, and we observed a significant decrease in surface FL-APP (p = 0.00446) (Fig. 4, F and G). Therefore, the observed accumulations of APP-CTF-positive puncta coupled with decreased surface FL-APP suggests that cyclopamine alters internalization and possibly retrograde trafficking that is required for γ-secretase-mediated cleavage of APP-CTFs.

Cyclopamine Alters Retrograde Trafficking and Promotes APP-CTFs Localization to Lysosomes

Previous reports indicate that upon endocytosis, FL-APP and APP-CTFs are localized to early endosomes and then sorted to either one of three possible trafficking pathways. One route is for FL-APP to be recycled back to the plasma membrane. For α- or β-secretase-cleaved APP fragments, APP-CTFs, a second route is available that allows these fragments to be retrogradely trafficked to the trans-Golgi network (TGN) for γ-secretase cleavage (19). Finally, APP-CTFs can be trafficked to the lysosome for degradation. To gain insight into these possibilities, we investigated the subcellular localization of APP-CTFs using immunofluorescence and subcellular fractionation.

To initially investigate where cyclopamine-induced APP-CTFs accumulate, we utilized coimmunofluorescence to identify the subcellular compartment(s) to which these APP-CTF puncta are localized (Figs. 5 and 6). Specifically, we assessed the markers of early endosomes (EEA1), late endosomes (MP6R), trans-Golgi network (TGN46), autophagosomes (LC3), and lysosomes (LAMP1) for accumulation of APP-CTFs. We initially noticed that upon cyclopamine treatment, the total intensity of EEA1, MP6R, LC3, and LAMP1 significantly increased (p = 1.18 × 10⁻¹¹, p = 2.52 × 10⁻⁶, p = 3.91 × 10⁻⁴², and p = 2.69 × 10⁻²², respectively), although no change was observed in TGN46 total intensity (Fig. 7, A–E). These data suggest that cyclopamine alters subcellular trafficking. In agreement with our findings, Jimenez-Sanchez et al. (55) recently showed that cyclopamine increases autophagosome formation.

FIGURE 5.

Endogenous APP-CTF distribution in vehicle-treated HeLa cells. Confocal three-dimensional analysis of HeLa cells treated with vehicle control (DMSO) for 24 h. Cells were stained for APP-CTFs using an APP C-terminal antibody (A8717) and antibodies (EEA1, MP6R, LAMP1, TGN46, and LC3) for subcellular markers (middle column). Right-hand columns are the merged images. Scale bar, 10 μm.

FIGURE 6.

Cyclopamine alters retrograde trafficking of APP-CTFs. Confocal three-dimensional analysis of HeLa cells treated with 5 μm cyclopamine for 24 h. Cells were stained for APP-CTFs using an APP C-terminal antibody (A8717) and antibodies (EEA1, MP6R, LAMP1, TGN46, and LC3) for subcellular markers (middle column). Right-hand columns are the merged images. Scale bar, 10 μm.

FIGURE 7.

Quantification of subcellular marker total intensity and APP-CTF subcellular localization in HeLa cells. A–E, single cells in each image were masked off, background intensities subtracted from each channel, and sum intensity of each cell measured and normalized to volume. Also, Manders' coefficients were calculated for each cell independently, and 26–51 cells were analyzed per experimental group. F–J, Manders' coefficients for colocalization. Dot plot diagrams represent raw data points, and the horizontal line represents the means of Manders' coefficients of subcellular marker and APP-CTF colocalization. Cyc, cyclopamine. Student's t test was used for statistical analysis: ***, p < 0.005; n.s., not significant.

To determine to which subcellular compartment APP-CTFs localizes, we quantified colocalization of APP-CTFs with these markers. Although the overall colocalization is low, we noticed that upon cyclopamine treatment there was a significant increase in APP colocalization with EEA1- MP6R-, and LAMP1-positive puncta (p = 1.17 × 10⁻⁶, p = 1.12 × 10⁻¹⁴, and p = 4.39 × 10⁻³³; respectively) (Fig. 7, F–H). There was a significant reduction in colocalization of APP-CTFs with TGN46 (p = 3.21 × 10⁻⁵) (Fig. 7I). No change in colocalization of APP-CTFs with LC3 was observed (Fig. 7J). In addition, we detected APP-CTF-positive puncta in close association with the ESCRT multivesicular body (MVB) markers, Tsg101 and Chmp2a (Fig. 8, A and B). These data suggest that cyclopamine decreases retrograde trafficking of APP-CTFs to the TGN while increasing trafficking to lysosomes.

FIGURE 8.

Immunofluorescence of APP-CTFs and MVBs in HeLa cells. Confocal three-dimensional analysis of HeLa cells treated with vehicle control (DMSO) (A) or 5 μm cyclopamine (B) for 24 h. Left panel, cells were costained using C-terminal APP (A8717) and an MVB ESCRT-I marker Tsg101. Middle panel, MVB ESCRT-III marker, Chmp2a-GFP, was overexpressed, and cells were stained with A8717. Right panel denotes all merged channels. Scale bar, 10 μm. As expected, APP-CTFs are in close association with MVB markers Tsg101 and Chmp2a-GFP upon cyclopamine treatment.

To independently verify that APP-CTF localization is altered upon cyclopamine treatment, we utilized subcellular fractionation of vehicle- and cyclopamine-treated HeLa cells (Fig. 9, A and B). Very modest changes in the distribution of subcellular markers such as EEA1 and LAMP1 were observed upon cyclopamine treatment. With respect to APP-CTF distribution, in vehicle-treated cells 69% of APP-CTFs are found in fractions 5–7, which partially overlap with the TGN marker (TGN46) (Fig. 9, C and H). However, in cyclopamine-treated cells we observed a shift in APP-CTF distribution; APP-CTFs in fractions 5–7 decreased to 37%, and an increase was observed in fractions 9 and 10. These latter fractions are enriched for the lysosomal marker LAMP1 (Fig. 9, C and G). In contrast to APP-CTFs, we did not detect an observable change in FL-APP distribution upon cyclopamine treatment (Fig. 9D). Thus, cyclopamine decreases trafficking of APP-CTFs to the TGN where γ-secretase-mediated Aβ generation occurs, and it increases APP-CTF transport to the lysosome. Because we did not observe a change in the distribution of PSEN1-CTF and FL-APP upon cyclopamine treatment, this suggests the effects are specific to APP-CTFs.

FIGURE 9.

Cyclopamine increases APP-CTF levels in lysosome-enriched compartments. HeLa cells treated with vehicle (DMSO) (A) or 5 μm cyclopamine (Cyc) (B) for 24 h were then collected and homogenized, and post-nuclear fractions were subjected to Optiprep step gradient fractionation. Fractions 2–13 (50–10% gradient) were subjected to Western blot analysis with APP C-terminal (c1/6.1) PSEN1-CTF, EEA1, LAMP1, and TGN46 antibodies. C–H, densitometry of each fraction as percent of combined total (fractions 2–13) densitometry for each respective protein.

Increased localization to lysosomes would suggest increased degradation of APP-CTFs. Surprisingly, we observed cyclopamine treatment increased APP-CTF levels. One way to rationalize our observations is that cyclopamine may attenuate lysosomal degradation of APP-CTFs.

Cyclopamine Decreases APP-CTF Lysosomal Degradation

To investigate whether cyclopamine affects APP-CTF degradation, we pretreated HeLa cells with cyclopamine for 24 h and then added cycloheximide to inhibit protein synthesis for an additional 0–4 h. At the end of this additional 4 h, APP-CTFs decreased by 76% in vehicle-treated cells, although only a 38% decrease was observed in cyclopamine-treated cells (p = 0.00218) (Fig. 10, A and B). Cyclopamine treatment nearly doubled the APP-CTF half-life from 2 to 3.6 h (Fig. 10B). Similar to our previous findings, cyclopamine did not alter the FL-APP rate of degradation (p = 0.0645) (Fig. 10C).

FIGURE 10.

Cyclopamine leads to moderate decrease in lysosomal maturation and significantly attenuates APP-CTF rate of lysosomal degradation. A, Western immunoblot analysis of FL-APP and APP-CTFs using a C-terminal APP antibody (c1/6.1). For FL-APP and APP-CTF analysis, naive HeLa cells were pretreated with 5 μm cyclopamine (Cyc) or vehicle (DMSO) for 24 h and then exposed to 50 μg/ml cycloheximide (CHX) for the indicated times. B and C, FL-APP and APP-CTF protein levels were normalized to β-actin first. The graphs represent protein levels as percent remaining of total protein at time 0 h. The lines represent the linear least squares fit where the slope of the line is the rate of protein degradation. D, naive HeLa cells treated with vehicle (DMSO) or 5 μm cyclopamine for 24 h followed by mature and immature cathepsin D detection in cell lysates via Western immunoblotting. E, bar diagram represents ratio of mature to immature cathepsin D protein quantification normalized to β-actin. Values denote mean ± S.E. Student's t test was used for statistical analysis as follows: **, p < 0.01; *, p < 0.05.

To test whether these APP-CTF degradation changes are due to decreased lysosomal maturation, we monitored cathepsin D levels, a reliable marker of lysosomal maturation (56, 57). We detected a modest but significant (p = 0.030) decrease in the ratio of mature (31 kDa) to immature (53 kDa) cathepsin D levels (Fig. 10, D and E). Together, our results implicate that cyclopamine leads to increased preferential retrograde trafficking of APP-CTFs to lysosomes and decreased lysosomal degradation of these APP-CTFs.

DISCUSSION

Here, we have discovered novel effects on APP trafficking and lysosomal maturation induced by treatment with cyclopamine. Specifically we demonstrate an accumulation of APP-CTFs in lysosomes and a decrease in Aβ and AICD generation.

After translation, APP is trafficked to the plasma membrane via the secretory pathway. APP can then be cleaved by α- or β-secretase at the plasma membrane or early endosome after endocytosis, respectively. These APP-CTFs are then retrogradely trafficked to the TGN for γ-secretase cleavage and Aβ generation (19). APP retrograde trafficking is highly regulated because APP proteolysis is dynamic and can lead to rapid changes in Aβ production (58). A consequence of decreased APP-CTF trafficking to the TGN is the decrease in γ-secretase-mediated cleavage of APP-CTFs and the concomitant decrease in Aβ and AICD generation. Hence, modulating APP retrograde trafficking independent of secretase activity can have novel implications for therapeutic avenues to treat AD.

Cyclopamine is a naturally occurring phytosterol isolated from the corn lily plant. The animal sterol, cholesterol, promotes amyloidogenic processing and increases Aβ generation, whereas cyclopamine exhibits the opposite effects on Aβ generation (29, 30). Phytosterols were recently shown to modify APP metabolism. Some phytosterols increased amyloidogenic processing and Aβ levels, and others decreased Aβ levels (32). In fact, stigmasterol demonstrated anti-amyloidogenic properties, decreased β-secretase cleavage of APP, and decreased Aβ generation. Moreover, mice fed stigmasterol-enriched diets showed decreased γ-secretase complex protein expression but lacked direct inhibition of γ-secretase activity in isolated mouse brain tissue (32). We also observed that cyclopamine treatment decreases γ-secretase-mediated proteolysis of APP without inhibiting γ-secretase activity directly. However, we observed cyclopamine treatment leads to the accumulation of APP-CTFs derived from α- and β-secretase equally. This is because the observed cyclopamine effects were downstream of APP-CTF generation at the plasma membrane (α-secretase) or early endosome (β-secretase).

Cyclopamine decreased APP-CTF retrograde trafficking to the TGN in HeLa cells. Choy et al. (19) demonstrated that γ-secretase cleavage of endocytic APP-CTFs and Aβ generation occurs at the TGN. Because cyclopamine does not inhibit γ-secretase activity, it is clear that the changes in APP-CTF levels are not due to changes in activity. Instead, these changes could be due to APP-CTF retrograde trafficking, thereby preventing colocalization with γ-secretase. Because we observed decrease localization of APP-CTFs at the TGN, we rationalize that altered trafficking is the mechanism by which cyclopamine decreases Aβ generation (Fig. 11). Alternatively, after APP-CTF endocytosis, these fragments can be trafficked to the late endosome/MVB and then to the lysosome for degradation. Here, we demonstrate cyclopamine increases APP-CTF trafficking to lysosomes. This increased trafficking to lysosomes could result in increased protein degradation and decreased APP-CTF levels. Surprisingly however, we observed increased APP-CTF levels and attenuated lysosomal degradation of APP-CTFs upon cyclopamine treatment. Interestingly, the accumulation of APP-CTFs was completely reversible upon washout of cyclopamine (data not shown). Therefore, it will be interesting to determine whether the APP-CTF accumulations are completely degraded after removal of the drug, and whether this acute drug exposure will still result in decreased Aβ levels. We observed similar changes in APP processing in HeLa cells and neurons. Our future studies will determine whether the changes in trafficking and lysosomal maturation observed in HeLa cells are also observed in neuronal cells.

FIGURE 11.

Model representation of APP-CTF retrograde trafficking and lysosomal localization upon cyclopamine exposure. Trafficking and cleavage of FL-APP and APP-CTFs in normal conditions (A) compared with cyclopamine treatment (B); FL-APP proteolysis and production of APP-CTFs occur at the plasma membrane (α-secretase) and early endosomes (β-secretase). APP-CTFs are then trafficked, via the retrograde pathway, to TGN for subsequent γ-secretase cleavage and Aβ generation (19). Alternatively, APP-CTFs are trafficked to late endosomes/multivesicular bodies thus destined for lysosomal degradation. B, cyclopamine treatment favors the lysosomal degradation trafficking pathway (bold arrows) of APP-CTFs thereby preventing γ-secretase proteolysis of APP-CTFs and Aβ generation.

Upon reaching the plasma membrane, FL-APP sheds the sAPP ectodomain. Lack of change in FL-APP and sAPP levels indicates cyclopamine does not alter the APP biosynthetic pathway. Decreased surface FL-APP suggests enhanced endocytosis after shedding, which can explain the increase in early endosome immunofluorescence intensity. The changes in Aβ, AICD, and APP-CTF levels in the absence of changes in FL-APP and sAPP levels imply that the effects of cyclopamine on APP metabolism are specific. The lack of change in FL-APP and sAPP levels upon cyclopamine treatment may also suggest it is a good candidate for possible AD therapy as it induces APP-CTF sequestration in lysosomes resulting in modest decreases in Aβ levels while not affecting FL-APP levels and sAPP generation. However, the chronic accumulation of APP-CTFs in the lysosome may be detrimental to protein homeostasis.

Both APP and Notch require primary cleavage at the plasma membrane for downstream endocytosis and retrograde trafficking to the TGN for γ-secretase and AICD/NICD generation (48, 59–61). Cyclopamine's effects may be specific to APP-CTFs because we did not see the exact same changes in exogenous ΔENotch processing. We did observe decreased NICD levels similar to the decrease in AICD levels. However, we did not observe increased ΔENotch levels as we did for exogenous APP-CTF-Gal4 and endogenous APP-CTFs. Because ΔENotch overexpression may result in nonphysiological ΔENotch trafficking and processing, we are reticent to conclude that cyclopamine's effects are specific to APP.

Cyclopamine is known to bind to and inhibit Smoothened (Smo) (62). Smo is a G protein-coupled receptor that is a component of the Sonic hedgehog (Shh) signaling pathway (63–65). It will be interesting to determine what role, if any, Smo and Shh play in APP trafficking and proteolysis. Given the ability of cyclopamine to decrease Aβ levels, it will be intriguing to determine whether cyclopamine is effective in treating AD transgenic animal models. One possible concern in utilizing cyclopamine is that it is a potent teratogen. The vast majority of AD patients are well beyond their reproductive age, which makes the teratogenicity less of a concern.

A balance between retrograde and lysosomal degradation trafficking pathways ensures proper distribution of APP and Notch holoproteins and their metabolites. Because these metabolites have been shown to be involved in regulating cell death/survival and synaptic plasticity, completely ablating the production of these metabolites would be detrimental (66–74). The modest but significant decrease observed in Aβ, AICD, and NICD levels suggests cyclopamine may not have the negative consequences that γ-secretase inhibition displays in some AD patients (75, 76).